Recently, I asked my LinkedIn network a simple question: “AI Adoption in GBS: What’s YOUR biggest barrier?” While 25% worried about proving ROI beyond pilots and 17% each cited scaling from proof‑of‑concept and stakeholder trust, the clear leader at 42% was: “Data isn’t ready.”
This aligns with Gartner research showing that 57% of organisations admit their data isn't AI‑ready. We’ve entered what Gartner calls the Trough of Disillusionment for Generative AI, where initial excitement meets operational complexity. Less than 30% of AI leaders say their CEOs are satisfied with AI investment returns.
But this is not an AI problem. It’s a data foundation problem. And that foundation is called Master Data Management (MDM), the critical enabler of AI readiness.
The Hidden Foundation of AI Success
Think about building a skyscraper. Before it reaches impressive heights or withstands storms, it needs deep, invisible foundations. The taller you want to build, the deeper those foundations must go.
Master data is your organisation’s hidden foundation. It determines how high your AI ambitions can reach and whether they'll withstand market volatility, regulatory change, and operational complexity. Yet most organisations are building AI skyscrapers on sand, investing in advanced AI while foundational data remains fragmented and inconsistent.
When AI projects fail, they often misdiagnose the issue as a technology problem instead of a foundation failure.
Master Data: The Cipher Key
Master data is the cipher key, the translation mechanism that allows everything else to be decoded and connected. Without it, your AI is trying to read encrypted messages without the key.
Consider a standard procure-to-pay process running across three systems: procurement, ERP, and finance. Your AI needs to understand the complete process. But when "Supplier ABC Ltd" in procurement is "ABC Limited" in ERP and "ABC Corp" in finance, your AI can't connect the dots.
It doesn't reject the bad data. It hallucinates connections. It invents relationships. It produces insights that look sophisticated but are fundamentally wrong.
This is why organizations are frustrated. Not because AI technology isn't ready, but because its data foundations can't support it.
The Good, The Bad, and The Ugly: A Diagnostic Framework
Understanding which category your data falls into is essential for assessing AI readiness.
The Good: AI-Ready Data
Clean, governed master data with a single source of truth and real-time quality monitoring. Ready for AI immediately.
Example: Standardized vendor master data with validated addresses, tax IDs, and banking information.
AI Outcome: High user trust drives adoption and ROI, competitive advantage.
The Bad: Risky for AI
Inconsistent but usable data with multiple versions of truth. Works for traditional reporting where humans apply judgment, but risky for AI.
Example: Customer data with duplicates and various naming conventions, but mostly complete.
AI Outcome: Users sceptical, manual overrides required, limited value, high maintenance costs.
The Ugly: Dangerous for AI
Siloed, contradictory, incomplete data with no governance or ownership. Dangerous for AI—produces wrong results confidently and at scale.
Example: Product data spread across systems with no synchronization, different coding schemes, and missing attributes.
AI Outcome: Loss of trust, project failure, wasted investment, damaged reputation.
Critical insight: AI amplifies whatever you feed it. Feed it "The Good," and you get a competitive advantage. Feed it "The Ugly," and you get competitive disaster—faster and at greater scale than ever possible before.
Three Dimensions of Data Readiness
This framework applies across three critical dimensions:
1. Data Quality Itself
Where does your master data sit on the spectrum? Customer data? Suppliers? Products? Employees?
2. Governance Approaches
Bad governance is worse than no governance—it adds cost without adding control. "Governance theatre"—500-page documents nobody reads, unenforced policies—creates the illusion of control while data quality deteriorates.
Good governance is invisible because it's built into how you work: clear ownership, accountability embedded in processes, enabling rather than blocking innovation.
3. What AI Does with Each
Same AI technology, completely different results:
The Good: Trusted systems, measurable ROI
The Bad: Skepticism, constant intervention, limited value
The Ugly: High-profile failures like Amazon's biased recruiting AI
The difference isn't the AI algorithm—it's the data foundation.
Proof Point: When Foundation Comes First
In a recent enterprise fraud‑prevention implementation, we consolidated approximately 475,000 vendor and bank account records across multiple systems down to 125,000 clean, verified records, eliminating 73% obsolete data. Only then did we implement AI‑powered external verification processes.
The project won industry recognition. But the key is this: it didn’t succeed because of superior AI tools. It succeeded because we built the data foundation first. Without clean master data, the same AI capabilities would have failed.
Organisations that invest in MDM before deploying AI see consistently better outcomes than those trying to “AI their way out of a data problem.”
Building the Foundation: A Practical Roadmap
Assess
- Rate your master data domains honestly: Good, Bad, or Ugly
- Review your top 3–5 AI use cases and the quality they require.
- Identify the most critical gaps.
Don’t sugarcoat. Ugly data doesn’t improve by calling it Bad.
Act
- Pick one critical dataset and get it to “Good.”
- Assign ownership and clear accountability.
- Embed governance into processes, not on top of them.
Prioritise the data that powers your highest‑value AI use case.
Accelerate
- Scale improvement domain by domain.
- Measure impact through business outcomes, not only data metrics.
- Make data quality a shared responsibility.
Demonstrate value quickly with early AI wins.
Key principle: You can’t AI your way out of a data problem, but you can data your way into AI success.
The Future of MDM in Autonomous Finance
As organisations move toward autonomous finance, MDM shifts from backoffice function to a strategic capability. AI can only analyse, decide, and take action reliably when master data is trustworthy.
Consider these emerging scenarios:
- Autonomous supplier onboarding, where AI evaluates risk and validates documentation.
- Real‑time fraud detection, which relies on consistent entity matching across systems.
- Intelligent contract analysis, where AI identifies terms, risks, and negotiation opportunities.
The AI capability already exists. What determines success is data quality, governance maturity, and organisational discipline, capabilities that take years to build and form a competitive moat.
The Strategic Imperative
The data readiness crisis is real: 42% of GBS leaders identify it as the biggest barrier to AI adoption. Master Data Management now sits at the core of every data strategy. The discipline has evolved into a strategic profession requiring a deep understanding of end-to-end data flows and governance that extends far beyond traditional boundaries.
Entering 2026, organisations with strong master data foundations will accelerate. Others will slow down, struggle, and explain away failed implementations. The question is no longer whether to invest in MDM; it’s whether you can afford not to.
Skyscrapers don’t fail because of what you can see. They fail because of what you can’t.
Mauro Portela is a Global Business Services leader specializing in master data management and business transformation. He publishes "Connecting.The.Dots," exploring data readiness, AI adoption, and the future of GBS.
Disclaimer: The content of this article is based on the author's own independent reflections and thoughts and is not in any way associated with any company or organization and has no commercial intent.